metadata
dataset_info:
features:
- name: preference_set
list:
- name: description
dtype: string
- name: dimension
dtype: string
- name: preference
dtype: string
- name: subdimension
dtype: string
- name: system
dtype: string
- name: prompt
dtype: string
- name: chosen
list:
- name: content
dtype: string
- name: role
dtype: string
- name: rejected
list:
- name: content
dtype: string
- name: role
dtype: string
- name: style_cluster_50
dtype: int32
- name: background_knowledge_cluster_50
dtype: int32
- name: informativeness_cluster_50
dtype: int32
- name: harmlessness_cluster_50
dtype: int32
- name: system_cluster_50
dtype: int32
splits:
- name: train
num_bytes: 619161257
num_examples: 65141
download_size: 338087967
dataset_size: 619161257
configs:
- config_name: default
data_files:
- split: train
path: data/train-*